仮説検証型研究における仮説の形式 ―主効果，調整，媒介，調整媒介についてのチュートリアル―

This article aims to clarify the common forms of hypotheses and explain how they are tested in industrial and organizational psychology, social psychology, and other disciplines. There are four types of testable hypotheses: main effect, moderation, mediation, and moderated mediation. Main effect hypotheses predict the relationship between two variables, for example, predictor X and outcome Y. Main effect provides a basis for the other three types of hypotheses. Moderation hypotheses focus on a variable, such as moderator W, which affects the association between two or more variables. Moderation hypotheses predict that the relationship between X and Y will change as a function of moderator W. Mediation hypotheses indicate the presence of an intervening variable, namely, mediator M that transmits the effect of X on Y. In other words, it is concerned with the indirect effect of X on Y through M. Finally, moderated mediation hypotheses refer to the notion that the indirect effect of X on Y through M is moderated by a fourth variable Z. These four types of hypotheses could be constructed from multi-level perspectives as well as a single-level perspective. Furthermore, how each type of hypothesis can be statistically tested is explained with reference to specific empirical studies. Finally, clarifying the form of the hypothesis will help reviewers and readers understand what researchers attempt to reveal. Such manuscripts can elicit a variety of comments from reviewers and readers, resulting in meaningful research publications.